Yasuhiro Fujita / 藤田康博

• Email: muupan@gmail.com
• Research interests: reinforcement learning, machine learning

Publications

Google Scholar

Peer-reviewed

• Surface-Aligned Neural Radiance Fields for Controllable 3D Human Synthesis
  • Tianhan Xu, Yasuhiro Fujita, Eiichi Matsumoto
  • CVPR, 2022. arXiv code
• ChainerRL: A Deep Reinforcement Learning Library
  • Yasuhiro Fujita, Prabhat Nagarajan, Toshiki Kataoka, Takahiro Ishikawa
  • Journal of Machine Learning Research, 22(77), 1-14. arXiv code (Chainer) code (PyTorch)
• Distributed Reinforcement Learning of Targeted Grasping with Active Vision for Mobile Manipulators
  • Yasuhiro Fujita, Kota Uenishi, Avinash Ummadisingu, Prabhat Nagarajan, Shimpei Masuda, Mario Ynocente Castro
  • IROS, 2020. arXiv
• Learning Latent State Spaces for Planning through Reward Prediction
  • Aaron Havens, Yi Ouyang, Prabhat Nagarajan, Yasuhiro Fujita
  • NeurIPS Deep Reinforcement Learning Workshop, 2019. arXiv
• A Wrapped Normal Distribution on Hyperbolic Space for Gradient-Based Learning
  • Yoshihiro Nagano, Shoichiro Yamaguchi, Yasuhiro Fujita, Masanori Koyama
  • ICML, 2019. arXiv
• Toward Onboard Control System for Mobile Robots via Deep Reinforcement Learning
  • Megumi Miyashita, Shirou Maruyama, Yasuhiro Fujita, Mitsuru Kusumoto, Tobias Pfeiffer, Eiichi Matsumoto, Ryosuke Okuta, Daisuke Okanohara
  • NeurIPS Deep Reinforcement Learning Workshop, 2018. pdf
• Model-Based Reinforcement Learning via Meta-Policy Optimization
  • Ignasi Clavera, Jonas Rothfuss, John Schulman, Yasuhiro Fujita, Tamim Asfour, Pieter Abbeel
  • CoRL, 2018. arXiv
• Clipped Action Policy Gradient
  • Yasuhiro Fujita, Shin-ichi Maeda
  • ICML, 2018. arXiv code slides

Non peer-reviewed

• Experience Replay with Random Reshuffling
  • Yasuhiro Fujita
  • Preprint, 2025. arXiv code
• Entropy Controllable Direct Preference Optimization
  • Motoki Omura, Yasuhiro Fujita, Toshiki Kataoka
  • Preprint, 2024. arXiv
• PLaMo-100B: A Ground-Up Language Model Designed for Japanese Proficiency
  • Preferred Elements, Inc.
  • Technical paper, 2024. arXiv

Translations

• Reinforcement Learning: An Introduction (second edition)
  • Co-translated into Japanese: 強化学習 (第2版)
• Algorithms of Reinforcement Learning
  • Co-translated into Japanese: 速習 強化学習 ―基礎理論とアルゴリズム―

Talks

• ゼロから始める深層強化学習 / Introduction of Deep Reinforcement Learning
  • 言語処理学会第24回年次大会(NLP2018)チュートリアル. slides

Code

• ChainerRL: A deep RL library in Python and Chainer
• PFRL: A deep RL library in Python and PyTorch
• async-rl: An A3C implementation in Python and Chainer
• DQN-in-the-Caffe: A DQN implementation in C++ and Caffe

Work experience

• Engineer at Preferred Elements, Inc. (November 2024 - Present)
  • Research and development in post-training of large language models.
• Engineer at Preferred Networks, Inc. (April 2015 - Present)
  • Research and development in machine learning for industrial applications: autonomous driving, robotics, computer graphics, and quantitative finance.
  • OSS development for reinforcement learning: ChainerRL and PFRL.

Social activities

• Program committee: Deep Reinforcement Learning Workshop at NeurIPS (2018-2022)
• Guest lecturer: RL part of 先端人工知能論II at the University of Tokyo (2016-2018)

Education

• M.S. Information Science and Technology (April 2013 - March 2015)
  • Graduate School of Information Science and Technology, The University of Tokyo
  • Thesis: “Automatic Feature Generation and Model Learning for General Game Players Based on Reinforcement Learning“
• B.S Engineering (April 2011 - March 2013)